Improve Plasmo build performance (dev in particular)

[tbrockman]

Example of a particularly terrible build (266834ms 😨):

🟡 0      | @plasmohq/parcel-transformer-manifest
🟡 1      | Adding icons
🟡 2      | Adding default_icon
🟡 3      | Adding default_popup
🟡 4      | Handling MV3 background service worker
🟡 5      | + Finished transforming manifest
Browserslist: caniuse-lite is outdated. Please run:
  npx update-browserslist-db@latest
  Why you should do it regularly: https://github.com/browserslist/update-db#readme
Browserslist: caniuse-lite is outdated. Please run:
  npx update-browserslist-db@latest
  Why you should do it regularly: https://github.com/browserslist/update-db#readme
Browserslist: caniuse-lite is outdated. Please run:
  npx update-browserslist-db@latest
  Why you should do it regularly: https://github.com/browserslist/update-db#readme
Browserslist: caniuse-lite is outdated. Please run:
  npx update-browserslist-db@latest
  Why you should do it regularly: https://github.com/browserslist/update-db#readme
🟡 0      | @plasmohq/parcel-bundler
🟡 0      | @plasmohq/parcel-runtime Injecting << page-runtime >> for popup.[hash].js 585f4e9c7d3dc21e D:\Local\otel-browser-extension\.plasmo\static\popup.tsx
🟡 1      | @plasmohq/parcel-runtime Injecting << background-service-runtime >> for static\background\index.js c338908e704c91f1 D:\Local\otel-browser-extension\.plasmo\static\background\index.ts
🟡 0      | @plasmohq/parcel-packager
🟡 0      | @plasmohq/parcel-optimizer-encapsulate static\background\index.js static\background\index.js
🟡 0      | @plasmohq/optimizer-es:  static\background\index.js static\background\index.js D:\Local\otel-browser-extension
🟡 1      | optimizer-es: skipped
🟡 0      | @plasmohq/parcel-optimizer-encapsulate popup.7d3dc21e.js popup.[hash].js
🟡 0      | @plasmohq/optimizer-es:  popup.7d3dc21e.js popup.[hash].js D:\Local\otel-browser-extension
🟡 1      | optimizer-es: skipped
🟢 DONE   | Extension re-packaged in 266834ms! 🚀

The norm seems to be somewhere in the seconds, which seems pretty excessive. Since we have the framework vendored already, should be pretty straightforward (hopefully), to dig into what process is taking so long during builds, and why.

Contribute any worthwhile changes or documentation upstream to the framework.